CorrectByConstructionTC-JVLC16

This repository documents the resources of the SHARE VM that belongs to the paper Roland Kluge, Michael Stein, Gergely Varró, Andy Schürr, Matthias Hollick, Max Mühlhäuser: "A Systematic Approach to Constructing Incremental Topology Control Algorithms Using Graph Transformation" DOI: http://dx.doi.org/10.1016/j.jvlc.2016.10.003 / URL: http://www.sciencedirect.com/science/article/pii/S1045926X15300483

VM metadata

• URL: http://is.ieis.tue.nl/staff/pvgorp/share/?page=ConfigureNewSession&vdi=XP-TUe_RK_CorrectByConstructionTC_JVLC.vdi
• User: sharevm[A}es[D}tu-darmstadt[D}de ([A} = @, [D} = ., [E} = !)
• Passsword: ShareVM[A}RealTime[E}

FAQs or What can I do with the VM?

Who can I contact?

If you experience problems, please do not hesistate to write a mail to Roland Kluge (roland[D}kluge[A}es[D}tu-darmstadt[D}de).

Where should I start?

1. Navigate to the Desktop (Win+D)
2. Double-click Eclipse Neon 1.a
3. All material is contained in the workspace that opens up.

Where is the metamodel? Where are the Topology Control rules?

1. In Eclipse, open the project de.tudarmstadt.maki.tc.cbctc.specification
2. Double-click the EAP file de.tudarmstadt.maki.cbctc.specification.eap
3. Enterprise Architect Lite opens up.
4. The metamodel can be found in the package de.tudarmstadt.maki.tc.cbctc.model. The Topology Control rules of kTC (and a number of other TC algorithms) can be found in the package de.tudarmstadt.maki.tc.cbctc.algorithms.

How can I run a sample simulation?

1. In eclipse, open the project simonstrator-simrunner
2. Navigate to *config/rkluge/launch/"
3. Right-click GUIRunner-with-EMF.launch and select Run as -> GUIRunner-with-EMF
4. Type the file name jvlc_complete_evaluation.xml into the search bar.

• This file contains all parameters of the simulation (e.g., energy model, movement model, protocol stack, overlay application).

5. Press Start Simulation
6. Two windows should pop up

• The Topology Visualization window shows the current output WiFi topology
• The Simulation window shows statistics about the progress of the simulation, simulation speed, etc.

7. Additionally, the log output in the Console view of Eclipse provides information about the progress of topology control.

What do the log messages tell me?

The Console view of Eclipse reports on Topology-Control-specific metrics. An example:

23-Nov-2016 17:06:46. INFO 	iter#001 CEs...
23-Nov-2016 17:06:46. INFO 	Adding node Node [id=1, properties={SiS-PHY_LOCATION=PositionVector [13.454846479205782, 73.0], SiS-LOCATION_X=13.454846479205782, SiS-LOCATION_Y=73.0}]
[...]
23-Nov-2016 17:06:46. INFO 	Adding node Node [id=101, properties={SiS-PHY_LOCATION=PositionVector [44.04043792076475, 106.56416757687579], SiS-LOCATION_X=44.04043792076475, SiS-LOCATION_Y=106.56416757687579}]
23-Nov-2016 17:06:46. INFO 	iter#001 CEs   Done (t=247ms, t_check=7ms, numCEs=[+n= 101, -n=   0, +e=1392, -e=   0, mod-d=   0, mod-w=   0, mod-p=   0])
23-Nov-2016 17:06:46. INFO 	iter#001 iTC...(algo=D_KTC(id=1), parameters=Parameters: {a=2.0, coneCount=4.0, k=1.41})
23-Nov-2016 17:06:47. INFO 	iter#001 iTC   Done (t=365ms,  t_check=89ms,  violations=0, allLSMs=1920 [[total count: a=1030, i=624, u= 266, effective counts: a= 790, i=602, u=   0]], effLSMs=1392)
23-Nov-2016 17:06:47. INFO 	iter#001 bTC...(algo=D_KTC(id=1), parameters=Parameters: {a=2.0, coneCount=4.0, k=1.41})
23-Nov-2016 17:06:47. INFO 	iter#001 bTC   Done (t=123ms,  t_check=115ms, violations=0, allLSMs=1392 [[total count: a= 790, i=602, u=   0, effective counts: a= 790, i=602, u=   0]], effLSMs=1392)
23-Nov-2016 17:06:47. INFO 	iter#001 iTC vs. bTC (CE+TC) wrt. [time=4.98 / LSMs=1.00]
23-Nov-2016 17:06:47. INFO 	iter#001 Stat..
23-Nov-2016 17:06:47. INFO 	Total real duration: 0.02min
23-Nov-2016 17:06:47. INFO 	iter#001 Stat  Done (duration=123ms, n=101, m=1812, avgBatPct=71.24, minBatPct=47.09, maxBatPct=96.27, numEmptyNodes=0, numSCCsInInput=1, numSCCsInOutput=1)
23-Nov-2016 17:06:47. INFO 	Simulated Realtime: 0:10:0:0 (H:m:s:ms)

Explanation:

• iter##001 Denotes the first iteration of the Topology Control algorithm. In this setup, Topology Control is executed periodically every 10min of simulated time.
• iter#001 CEs... Adding node Node [...] Before running the algorithm, all pending context events are handled.
• iter#001 CEs Done (t=247ms, t_check=7ms, violation=0,... At the end of context event handling, the required CPU time for the handling and for the constraint checking is printed. Additionally, the number of detected constraint violations is reported.
• iter#001 iTC...(algo=D_KTC(id=1), parameters=Parameters: {a=2.0, coneCount=4.0, k=1.41}) Then, incremental the topology control algorithm is invoked (here: the traditional kTC with k=1.41). Afterwards, the required CPU time for the algorithm and the constraint checking are printed.
• iter#001 bTC...(algo=D_KTC(id=1), parameters=Parameters: {a=2.0, coneCount=4.0, k=1.41}) For comparison reasons, the batch counterpart of the current algorithm is executed on a copy of the topology and runtime statistics are printed.
• iter#001 iTC vs. bTC (CE+TC) wrt. [time=4.98 / LSMs=1.00] At the end of a Topology Control run, the costs of the incremental and batch algorithm are compared in terms of CPU time and link state modifications. In this case, the incremental algorithm needed ca. 5 times longer compare to the batch algorithm. The link state modification count is identical in this iteration because the topology is processed for the first time.

Concerning the research questions

Unfortunately, the SHARE VM is not powerful enough to reproduce the full evaluation, however you can get an idea of the how the results were produced.

• RQ1-Correctness After handling context events and invoking Topology Control, the per-algorithm constraints are checked and violations are reported (see violations=0 above).
• RQ2-Incrementality The comparison iTC vs. bTC summarizes the incrementality per Topology Control iteration.
• RQ3-Performance The CPU time and the link state modification count is recorded for all major steps of a Topology Control run (e.g., t=220ms, t_check=11ms).
• RQ4-Generality This is a research question that is discussed extensively in the paper. Still, the number of implemented Topology Control algorithms indicates the applicability of the approach.

There is also a class that runs the entire evaluation: JVLCEvaluationExecutor (search for it via Ctrl+Shift+R) This class produces files that can then be processed using R:

1. Start the launch configuration simonstrator-simrunner/config/rkluge/jvlc/Rconsole.launch.